Integrating management practices into river health assessment: A cloud model framework for small and medium rivers under uncertainty

Ten small and medium rivers (SMRs) in Gaochun District, Nanjing, China. This study develops an integrated river health assessment framework for SMRs that incorporates management-related indicators from a ‘lifestyle’ perspective alongside conventional biophysical metrics, covering six dimensions of river morphology, riparian condition, hydrology and water quality, aquatic life, social services, and river management. To capture the uncertainty arising from hydrological variability, data fluctuations, fuzzy classification boundaries, and inter-indicator coupling, a hyper-entropy expectation curve cluster is incorporated into the multi-indicator cloud similarity model for health classification. SHapley Additive exPlanations (SHAP) identify the importance of indicators and key drivers of river health variations. Rivers in Gaochun District are generally classified as ‘sub-healthy’. Overall, 71.4% of indicators reach at least a ‘healthy’ level, with basic operational safety ensured and water quality effectively controlled. However, aquatic life and river management remain at an ‘unhealthy’ level, constraining overall river health. Specifically, simplified habitat structure and insufficient flow-velocity heterogeneity limit habitat diversity and biological recovery, while unclear management boundaries and limited integrated management capacity further restrict sustained improvement. SHAP analysis indicates that management-related indicators contribute about 32% of the total importance, ranking as the dominant contributor and highlighting the critical role of management in shaping regional river health patterns. These findings underscore the need for targeted freshwater management and ecological restoration. • Management practices were incorporated into river health assessment. • An improved multidimensional cloud similarity model was developed to better capture fuzziness, randomness, and uncertainty. • A composite weighting approach balanced subjective judgment, objective data, and inter-indicator interactions. • SHAP analysis identified key drivers of river health and enhanced model interpretability.